Transcription
Hello everybody and welcome back. We have finally made it through the three major pulse sequences: spin echo, gradient echo, and inversion recovery. I want to spend some time now looking at some of these subtleties or nuances that can be changed within these pulse sequences and then move on to looking at more advanced MRI pulse sequences that you're going to come across in your MRI physics studying.
I've decided to start off by looking at this concept of saturation. Now, saturation is an incredibly important concept to understand as we head into these more advanced techniques. When we focus on MR angiography and time of flight, we'll see that saturation plays an important role. When we look at trying to reduce artifacts, a motion artifact within an image, we use something known as saturation bands to reduce artifacts, and understanding the concept of saturation is incredibly important within these pulse sequences.
Now, today I'm going to focus on a specific saturation technique known as CHESS, or chemical shift selective saturation sequences. Now, you'll see that I've separated this into partial saturation and saturation techniques, and these are fundamentally different from one another, and we need to know the difference between the two.
Now, partial saturation is something we've actually come across before, we just haven't formally named it. When we looked at the Ernst angle, the angle that was providing the maximum signal for a specific tissue at a specific TR, we saw that at angles higher than that Ernst angle, we started to lose signal. We got this drop-off of signal. What was happening here is what's known as partial saturation. When we flip the transverse magnetization vector by a specific angle, we get a certain degree of transverse magnetization. The Ernst angle is the angle that we can flip that net magnetization vector in that specific tissue that will allow for full recovery of longitudinal magnetization before the next TR. At any angles higher than this, we won't get full recovery of longitudinal magnetization before the next TR, and we know that at the TR, at the RF pulse that we apply to flip that spin again, or that net magnetization vector again, it's the degree of longitudinal recovery that's going to determine the magnitude of that transverse magnetization. And if we're not fully recovering that longitudinal magnetization, we're going to get a drop-off in the degree of transverse magnetization, and that's why at higher flip angles, we were getting a drop-off of signal here. That net magnetization vector within that specific tissue is said to have partial saturation. There's been loss of signal because it hasn't been allowed to fully recover its longitudinal magnetization.
Now, we can look at this in a different way by looking at the changes in longitudinal magnetization over time for two separate tissues. Here, we're going to look at fat and CSF. Now, prior to the first 90-degree RF pulse, if we're using 90 degrees in this case, both fat and CSF will have full longitudinal magnetization. They are fully relaxed in the B plane, in the longitudinal plane, at one point. Then we are going to flip the net magnetization vectors for each of these to 90 degrees. Both fat and CSF have now been flipped to 90 degrees, and we're getting maximum signal because we had maximal longitudinal recovery. That first flip is going to cause the longitudinal magnetization to go to zero. We lose all of the longitudinal magnetization, and we have maximum signal here in the transverse plane. Hopefully, you can see it's the degree of longitudinal magnetization that is going to determine the degree of transverse magnetization. So, we can see that the changes in longitudinal magnetization are going to correspond to the transverse magnetization at the next RF pulse.
Now, what happens if we allow these two tissues to fully relax over time? What's going to happen to that longitudinal magnetization? We've seen here, we've lost all of our longitudinal magnetization, and then we're going to get recovery of longitudinal magnetization at different rates for the different tissues depending on their T1 values, their spin-lattice interaction causing longitudinal recovery. So, if we plot that on a graph, we can see that fat recovers its longitudinal magnetization faster than CSF does. There is more spin-lattice interaction in fat than there is in CSF. Now, that quicker recovery of fat shows us that fat has a shorter T1 time or T1 constant than CSF does. If then we were to do another RF pulse here at this point in time, we would see that both the tissues have regained their longitudinal magnetization. We would flip them again 90 degrees, and we would have maximum signal again. Our signal here would correspond to this longitudinal recovery.
Now, what happens if we reduce that TR time down? We flip the spins prior to them reaching their full longitudinal recovery. Here, I've given you an example where we have made the TR a quarter of what it was initially. We are now flipping these spins at 90 degrees at each one of these dotted lines here, and I want to represent what's happening on these Cartesian planes here that match up with these shorter TRs.
Now, what happens to these vectors as we wait till this TR? Well, we've seen that fat is going to recover its longitudinal magnetization faster than the CSF has. So, here we can see that fat has recovered some longitudinal magnetization to this level here. CSF has recovered less longitudinal magnetization because it has a longer T1 time. Now, we know that if we were to flip these now to 90 degrees, the vector, the transverse vector for fat, is going to be much higher than the transverse vector for CSF because it's the longitudinal magnetization, or the longitudinal recovery, that determines the degree of transverse magnetization at the next 90-degree RF pulse at the TR here. Now, you can see that the difference is represented here in this graph, and this is the T1 contrast differences that we get between tissues at short TR values.
If then we were to flip these again to 90 degrees, we would lose longitudinal magnetization completely. Remember, in this period of time, we've fully lost transverse signal. Free induction decay happens much more rapidly. This vector here is only representing the longitudinal magnetization. It's not representing the transverse magnetization. We get rapid loss of transverse magnetization with spin-spin interaction, and that transverse magnetization gets lost much before we get recovery of the longitudinal magnetization vector. So, these arrows here are only representing the degree of longitudinal magnetization that we're recovering. Again, we lose longitudinal magnetization here because we flipped it to 90 degrees, and then they will recover at the same rate as they did the T1 for those specific tissues again.
Now, our vectors are going to be different at this point. The longitudinal magnetization that we had at this point here will be flipped into the transverse plane, and it will recover during this TR interval here. That recovery will give us a different longitudinal magnetization vector. For CSF, we only had a very small amount of longitudinal recovery. At this 90-degree flip, we're only going to have a small transverse vector. That small transverse vector then regains longitudinal magnetization at the same rate that this vector did, the T1 for CSF again. It's going to give us a longitudinal magnetization vector that's ever so slightly less than our initial longitudinal magnetization vector.
This process can then repeat itself with successive TRs, and we see that what we start to get is a steady-state signal between the two tissues. The degree of longitudinal relaxation is equal with every successive TR. We've reached a steady state where our flip angle and the degree of longitudinal magnetization is the same for each and every successive TR, and we can then plot these out and see that different tissues, based on the T1 of that tissue and the length of the TR, is going to come to a different steady state. And we can see that after three 90-degree RF pulses, we are at that steady state. This is what's known as partial saturation. These tissues are still giving off signal, but that signal is less than that initial 90-degree RF pulse. They're said to be partially saturated. The difference between this signal here and the T1, if we were to wait as a long TR, that difference is the partial saturation difference.
Here, this is not the same as saturation. Partial saturation is something that naturally occurs based on the TR that we've selected for our pulse sequence. Saturation is when we specifically try to get rid of signal either from a specific tissue or from a specific region or from tissue that is entering our slice during the pulse sequence, and we're going to look at those differences in upcoming talks. Today, I'm going to focus on how we can select a specific tissue and use a saturation technique, which is an active technique that we are using to try and get signal to be lost for that specific tissue. Partial saturation isn't an active technique; it's a fact of life. It's what's happening during our pulse sequences, and we can't get around that based on the TR that we selected.
Let's look at what exactly saturation is and how that differs from partial saturation again. Saturation is something that we are actively doing. We have added elements to our sequence in order to null signal from either a specific tissue, a specific region, or a tissue that's entering or exiting our slice. So, what we do here is we have net magnetization vectors in the longitudinal plane. We've got two separate vectors here. They're lying within the longitudinal plane. They are along the B0 of our magnet. Now, what saturation does is we can apply a specific pulse that selects either a specific tissue or a specific region or a tissue that's entering or exiting a slice. This RF pulse selects a specific tissue to move into the transverse plane without moving the other tissues into the transverse plane. It's only selecting this tissue.
Now, what happens is the tissue that we've selected then goes into the transverse plane and theoretically would provide some signal. We then apply what's known as a spoiler gradient. Spoiler gradients are a combination of frequency encoding and phase encoding gradients that are applied to dephase completely the signal in the transverse plane. Anything that we flipped into the transverse plane, we want those to completely lose phase, to completely lose signal. As we apply that spoiler gradient, we can see that we get phase loss in the transverse plane. The tissue that didn't get flipped is still just processing along the main bore of our magnet. It's being unaffected by this radio frequency pulse, all these spoiler gradients.
Now that we have lost transverse magnetization, the net vector for these spins that we flipped into the 90-degree plane is zero in the longitudinal plane. They completely lost their longitudinal magnetization, and they've completely lost their transverse magnetization because they are completely out of phase with one another. We have now got no longitudinal or transverse magnetization for the specific tissue that we selected. Then, if we go about doing our normal pulse sequence here, I've included a spin echo pulse sequence. It could be a gradient echo pulse sequence. If we do our normal pulse sequence, it's only going to flip the longitudinal vector into the transverse plane. We've seen it's the longitudinal vector that determines how much gets placed into the transverse plane. Only the longitudinal vector that remains is going to contribute to this pulse sequence here. It's going to flip into the transverse plane, it's going to provide our signal at the TE, it's going to relax until our TR until we repeat, repeat the process again.
Now, importantly, the way I've drawn this shows large gaps between this initial RF pulse and our sequence. In reality, these are very closely spaced. We don't want these spins to regain any longitudinal magnetization prior to our sequence happening, so these are tightly packed prior to our pulse sequence.
Now, how exactly do we go about selecting only a specific tissue to flip with this initial RF pulse here? Well, you see I've drawn this RF pulse with a different color here. The bandwidth of this RF pulse is different from the bandwidth of the RF pulse that we're going to use in our sequence. I want to use a specific example here to show you how we can spectrally select and saturate a specific tissue. Later on, we're going to look at how we can spatially select or how we can select based on flow within tissues. This is purely an example, and it is what is known as spectral saturation or chemical shift selective saturation techniques.
Now, we've seen previously that water and fat have what's known as chemical shift. They've got slight precessional frequency differences in their hydrogen atoms. We've seen that the Larmor frequency, the frequency at which a spin precesses, is based on the gyromagnetic ratio and the local magnetic field experienced by the hydrogen protons within those atoms that we're looking at or within those molecules that we're looking at. And we've seen in water that the oxygen atom descreens the hydrogen atoms within the water molecule by keeping the electrons more likely to be around the oxygen than they are around the hydrogen. The local magnetic field that this hydrogen experiences then is high. In fat, because of the molecular structure of fat, we get what's known as shielding, where the electrons in this molecular structure shield the hydrogens from some of the external magnetic field, and it's that slight difference in local magnetic field that causes precessional frequency differences between water and between fat. We can see that water precesses at a frequency that is ever so slightly higher than fat within our magnetic field, and the difference between these two frequencies is what's known as the chemical shift. And with water and fat, we've seen that this chemical shift is 3.5 parts per million.
Now, what does that mean? Well, for every million Hertz that water is precessing, fat will precess at 3.5 Hertz less than water. It's a minute difference between the two, and that means at 1.5 Tesla, water, we know, will precess at 64 million Hertz. Water's precessing at 64 million Hertz. For every million Hertz that water processes, fat is going to process at 3.5 Hertz less than water. So, for 64 million Hertz, we'll see that fat precesses at 224 Hertz less than water. Water. Now, why is this important? This frequency difference here is 224 Hertz in a 1.5 Tesla machine. Let's go back to our pulse sequence here and see how we can utilize that. We've got water and we've got fat precessing at roughly 64 million Hertz, but we know that fat is precessing at 224 Hertz less than the water is.
This RF pulse will only flip spins into the transverse plane that match the precessional frequency with the RF pulse. That's the basis for slice selection. When we apply an RF pulse, that RF pulse needs to match the precessional frequency in order for resonance to occur. What then would happen if we applied an RF pulse that was very narrow in bandwidth? It only selected a very small bandwidth, a specific precessional frequency, and we made sure that that precessional frequency matched fat and fat alone. That RF pulse was so accurate that it only matched the precessional frequency of fat and it didn't match that precessional frequency of water. Remember, only 224 Hertz difference needs to be a very accurate RF pulse. Doing so with that narrow RF pulse would allow only fat to be flipped, only fat to resonate in phase with one another, flipped into the transverse plane. Water would remain precessing in the B0 for longitudinal direction because it didn't match this RF pulse. That is how we specifically select for fat based on fat's precessional frequency.
In inversion recovery, the way we null signal coming from fat was by using a time to inversion that matched the T1 recovery rate of fat. It wasn't specific for fat; it was specific for that T1 recovery rate. Here, we are specifically selecting for fat because we are matching the exact precessional frequency of the hydrogen within fat within this magnetic field. Notice how we haven't slice selected here, so this is happening within the entire bore of the scanner. Remember, when we slice select, we apply a gradient in the longitudinal direction that will cause differences in precessional frequencies along the longitudinal axis of the patient. That's not what we want. We want to accurately select all of the fat within the patient by applying an RF pulse that matches the precessional frequency of fat. When just the main magnetic field is on here, the only magnetic field that the spins are experiencing is the main magnetic field. The fat then is flipped into the transverse plane, and we apply these spoiler gradients. This all happens very close to our pulse sequence. Those spoiler gradients ensure the fat has no transverse magnetization and it has no longitudinal magnetization.
Now, when we apply our slice-selected radio frequency pulse, only the spins that match the precessional frequency of that radio frequency pulse and only the spins that have longitudinal magnetization will then be included within this pulse sequence here, and that's how we can suppress signal coming specifically from fat. Now, you'll notice that this 90-degree RF pulse was the same 90-degree RF pulse that we've used throughout our pulse sequences. You might be thinking, why can't we just use a 90-degree RF pulse that specifically selects for water here? If we were to do that, that's a great idea. If we were to do that, only the water would then flip, and we could just do our pulse sequence with a 90-degree RF pulse that was matched to water. The bandwidth of our radio frequency pulse determines the slice thickness. If that doesn't make sense to you, go back to slice selection and see how radio frequency bandwidth affects slice thickness. If we were to select water only with this 90-degree RF pulse, make it a very narrow RF bandwidth, we're going to get an incredibly thin slice that we are imaging, a slice that's not going to give us enough signal for that image. We're going to get a very poor signal-to-noise ratio. Nulling fat first, preventing fat from contributing to signal within our image, allows us to run a normal pulse sequence with a wide bandwidth. It doesn't matter if we select frequencies that fat would have had because fat is no longer contributing to signal. We can select the appropriate thickness slice and create the image as we've discussed in spin echo or gradient echo or inversion recovery pulse sequences.
So, now we've looked at two ways to suppress signal from fat. We've looked at inversion recovery using a T1, and we've looked at chemical shift. Now, inversion recovery had the problem that if we used contrast, gadolinium-based contrast within our image, that gadolinium is going to reduce the T1 times of the tissues to a level that's similar to fat, and we're going to null the signal coming from gadolinium, which is not what we want. Here, in chemical shift selective sequences, we don't have that problem. The gadolinium changes the T1 of the tissue, but it doesn't change the precessional frequency of the tissue. So, we can still null signals coming from fat and still get signal coming from gadolinium that's in tissues. So, this is a great type of sequence if we're using gadolinium contrast and we want to get rid of signal coming from fat.
Now, it's not all good news when it comes to chemical shift, because as you can see, we need to be very accurate in the precessional frequencies that we're selecting with this 90-degree RF pulse. If there are inhomogeneities in our field, if our machine is not giving a very accurate B0, we are not going to accurately be able to select fat. If there are local magnetic field inhomogeneities, say there's metal hardware within the image, that metal is going to change the local environment, the local magnetic field, and we're not going to be able to suppress signal coming from fat in those instances. Perhaps STIR is a better sequence to use. There are trade-offs that come with the different sequences that null signal coming from fat.
Now, this is a great question that comes up in exams. There are actually three main mechanisms to null signal coming from fat, and we go through that extensively within the question bank that's linked below. We are still going to look at the last mechanism within these lectures of reducing signal coming from fat, that's coming up in a later talk.
Now, we're going to move on to what's known as fast imaging or rapid imaging, and I'm going to show you how we can manipulate spin echo and gradient echo sequences to create even faster pulse sequences for acquiring our images. There are certain instances where we need to acquire an MRI sequence quickly, and it's these rapid sequences that allow us to do so. So, until that talk, I'll see you there. Goodbye.